Using Simulations to Answer Cropping Questions on Dairy Farms
Greg Roth
Department of Crop and Soil Sciences
Penn State
September 2001
Making cropping system recommendations on dairy operations is complex. Many agronomic decisions can impact feed production, feed quality, nutrient management, milk production and profitability. On top of that our variability in weather from year to year makes assessing the best option difficult over the long term. Using a computer simulation of a dairy farm might be one alternative to approaching these problems.
During the last year I have worked with Dr. Al Rotz at the Pasture Systems and Watershed Research lab and some our colleagues to use Al's dairy farm simulation model called DAFOSYM (http://www.ars.usda.gov/Main/docs.htm?docid=8518) to address several crop production issues on dairy farms. I have been reasonably impressed with the results.
Basically, the model simulates the performance of a dairy farm under several management alternatives. The user enters all of the characteristics of the farm- soils, cows, crops, and equipment. The program then simulates the growth of the crops, the timeliness of harvest, the forage quality of the forages, balances the ration, and then estimates milk production, nutrient losses and the economic performance of the farm. It does this over a 30-year period using log term weather data. You can then adjust some of the predicted results to match records of actual farm performance.
In the first part of this effort we found the corn and soybean yield models seemed to predict actual yields fairly well. The model gives you a prediction of all crop yields for the last 30 years.
We conducted several simulation projects during the last year. In our first trial, we looked at the economics of adding soybeans to a corn/alfalfa rotation in Centre or Franklin County.
We found that soybeans are a profitable addition on farms that have excess land available for grain production and low corn prices. They weren't profitable when you needed all the corn for silage. The model also suggested that soybeans were more valuable when sold as a cash crop rather than fed on the farm. Soybeans had little impact on environmental benefits.
We have done other simulations with adding additional acres, adding cover crops and substituting small grains for corn. As with all models, the results must be evaluated carefully but it does allow you to test possible scenarios without doing extensive field research.
I'm excited about working with the model more because its difficult to do cropping system studies when you have to measure actual yields. Often actual yields vary considerably from year to year and field to field and then you need lots of years or locations to say anything meaningful.
Eventually we should also be able to do simulations of different grain crop rotations for grain producers. That should be interesting as well.